EBS : Elastic Block Storage, one of the web services that we believe is fundamental for a self-respecting cloud computing infrastructure, now allows booting.
What is EBS: A service capable of creating a storage volume that we can link to an EC2 server instance, is defined as persistent to differentiate it from the storage with which the EC2 virtual instance is born, which is not persistent as the shutdown of the same port with it the loss of data and all configurations.
By drawing parallels with VMware’s virtual infrastructure, we can associate EC2 instances with virtual machines defined on a hypervisor with internal disks, while EBS is storage external to the hypervisor, where we can define LUNs and make them visible to our virtual machine. The virtual machine will see the new disk as a local disk.
This new disk is more performing, it resides in a storage where there is a RAID policy (which is why it is persistent) and on which we can also take snapshots (on S3) from which we can produce replicas of our data as well as migrate them between AWS datacenters.
Let’s summarize its history:
21-08-2008
– The release of this fundamental feature is announced.
10-12-2008
– The possibility of using the Elastic Compute Cloud (EC2) and therefore also the EBS feature is extended to European datacenters.
24-09-2009
– The possibility of sharing snapshots between co-workers, i.e. other AWS users, is announced.
04-12-2009
– The ability to boot directly from the EBS is announced.
This new feature is the big step of AWS services, in this way Amazon’s Cloud Computing infrastructure recovers a very important gap that could have made it lose a lot compared to all the new competitors who are brutally entering and establishing themselves.
Still wanting to make a parallel with the VMware virtual infrastructure, the virtual machines instantiated on the hypervisor with the virtual disks residing on the hypervisor, are subject to non-persistence and non-reliability (see the AWS AMIs instantiated by S3), while if the image of the virtual machine is resident (including boot disk) on a shared storage and is simply instantiated on the hypervisor, it benefits from the persistence and speed of start-up, as well as the performance of the storage system (see AMIs instantiated by the EBS).
Maybe soon we will have an HA (High Availability) service or maybe directly an FT (Fault Tolerance) service, but the possibility of being able to share the EBS disk between two AMIs, like a classic Cluster between two servers, is still missing.